How AIOStory measures, and what it refuses to make up
AIOStory reads the locations you enter from public on-page signals, scores each on six pillars with the AIOTruth evaluation engine, and compares the results: into a six-chapter story across all of a brand's locations, or two of its locations one against the other. Every sentence traces to a measured value. No LLM in the scoring or synthesis. No randomness. The same inputs always produce the same result, word for word.
Evaluate, compare, present, act
AIOTruth evaluates the public signals on each site. AIOStory compares the results and reveals where the story stays consistent, changes, or breaks. AIOInsights brings the systems together as the front door. Digilu takes responsibility for what happens next.
So the division is clean: AIOInsights is where a single site is evaluated, powered by AIOTruth. Comparison, one brand across all its locations or two of its locations one against the other, is powered by AIOStory. Definitions and standards live at AIOFacts, and change over time at AIOWeather. Full strategy, implementation, and ongoing work are Digilu.
What is retrieved, scored, compared, and what is not
Retrieved: each site's public homepage-level signals, plus robots.txt, sitemap.xml, and llms.txt, fetched the way an AI crawler would.
Scored: six pillars per site, from observable on-page evidence only, on the AIOTruth 0 to 10 scale.
Compared: those per-location results, either across all of a brand's locations or between two of them, into the visible output.
Synthesized: the six-chapter Location Story, through the fixed pillars-to-chapters.json mapping.
Not measured: we do not query ChatGPT, Claude, Gemini, Perplexity, or AI Overviews, and a deterministic on-page scan is never presented as what one of those platforms recommends. No LLM sits in the scoring or the comparison.
Storage: a Site Comparison is a private read for you. Nothing about it is published, listed, or emailed, and the other business is not notified. Every result carries its evaluation date and rubric version.
Blocked or unreachable: a site that blocks AI crawlers or cannot be reached returns an honest status, never a guessed score. A change to a site's public pages is the only thing that changes its result.
Six pillars, scored per location
AIOStory does not invent its own rubric. Each site is scored on the same six-pillar AIOTruth framework that powers the AIOInsights single-site check. In Site Comparison, AIOStory calls the same AIOInsights evaluation endpoint the single-site check uses, so a site's score here matches its own AIOInsights check exactly. Each pillar is scored 0 to 10, shown as 0 to 100. A signal is strong at 70 and above, weak below 50.
Clarity
Can AI tell what you do.
Identity
Does AI know who you are.
Substance
Is there enough to quote.
Reach
Can AI read the page at all.
Trust
Can AI see proof you are trusted.
Locality
Can AI place you on the map.
Six chapters, every line traced to data
The per-location scores feed a fixed synthesis that produces six story chapters. Which pillars feed which chapter is not improvised at runtime. The mapping is defined in a fixed file, pillars-to-chapters.json, so chapter feeding is deterministic and auditable.
The six chapters
Who AI Thinks Your Brand Is reports the dominant brand name AI reads and how confidently it can describe the brand, from Clarity, Identity, and Substance averaged across readable locations.
What AI Gets Wrong surfaces on-page inconsistencies: more than one brand name across locations, or wide swings in how clearly each location states who it is. This is on-page consistency only. Factual freshness of hours, addresses, and offers is checked per location in a custom full evaluation from Digilu, not here.
Where AI Looks Elsewhere lists locations AI cannot read at all, blocked or unreachable, or barely reads with low Reach. When AI cannot read a location, it recommends one it can.
Missing Chapters flags any pillar that scores weak at most readable locations. That is treated as a brand-level gap, not a one-location fix.
Trust Signals reports which locations expose machine-readable proof, ratings in structured data, linked review profiles, local-business signals, and which do not.
The Fuller Story states honestly how many locations were sampled, N of M, and what a custom full evaluation from Digilu adds.
If a chapter has no data behind it, the report says "Not enough sampled locations to assess" rather than inventing filler.
A corporate site plus three locations, and N of M is always disclosed
The free read takes your corporate site plus up to three locations, four URLs in total (SAMPLE_CAP = 4). We dedupe by host first, so the same site entered twice counts once. The report always discloses the coverage as "N of M sampled" so you know how much of your brand was read.
A custom full evaluation from Digilu, the paid step, reads every location and microsite individually, or your full competitor set. It is the only path that goes beyond this four URL sample and beyond homepage-level signals, and it is what you contact Digilu for when you run more than a handful of locations.
When AI cannot read a website
Some locations are invisible to AI. AIOStory detects this directly. A location is marked unreadable when it is blocked by a bot challenge, or when it is unreachable. Those locations are not scored on the six pillars as if everything were fine. They appear in the "Where AI Looks Elsewhere" chapter, because when AI cannot read one location it will recommend a location it can.
Deterministic, and nothing invented
There is no LLM in the scoring or the synthesis. There is no randomness. There is no paid API call to a chatbot. The pillar scores come from the AIOTruth evaluation engine, and the chapter mapping is the fixed pillars-to-chapters.json file. As a result, the same inputs yield the same story word for word.
We never claim anything we do not measure. Every chapter sentence traces back to a value we read. When the data is not there, we say so instead of filling the gap.
What the free read does not do
We read homepage-level public signals per location. We do not crawl every page of every site in the free read.
Factual freshness, whether hours, addresses, and offers are current and consistent, and per-microsite depth are full-audit work only. The free read does not verify them.
Reviews are read from on-site machine-readable signals, structured data and linked profiles exposed on the location's own site. Reviews living only on Google or Yelp do not count unless they are exposed on the site where AI can read them. We do not query live Google or Yelp.
See how consistently AI systems can understand your locations
It takes your location URLs and about two minutes. The story might surprise you.
Reveal my brand story